summaryrefslogtreecommitdiff
AgeCommit message (Collapse)AuthorFilesLines
2007-10-10r5284: this is needed on my serverStefan Metzmacher1-0/+2
+/* this #undef is needed on my SuSE 9.2 Box with glibc-devel-2.3.3-118 */ +#undef u32 as sys/epoll.h has this: typedef union epoll_data { void *ptr; int fd; uint32_t u32; uint64_t u64; } epoll_data_t; metze (This used to be commit b4928f3ea8f6530317dfa76593cf93e6acd4337a)
2007-10-10r5282: merging autogen fixes from 3_0Gerald Carter1-3/+3
(This used to be commit ece7a3918ec33b4b8b61b50277c5a1e484c55e3b)
2007-10-10r5277: initialise the multi_homed flag in the name registration testAndrew Tridgell1-0/+1
(This used to be commit d51fcee29a8d8bf253b3a6d0b7b741f9b55af94a)
2007-10-10r5276: - added support for NBT_OPCODE_REFRESH2 (type 0x9)Andrew Tridgell5-34/+145
- when registering with a WINS server, initially use multi-homed registration, then switch to name refresh requests. Send refresh requests only to the WINS server that responded to our registration. If that server goes away, then start the registration from scratch. This makes registration more robust to WINS server failure. - send WINS registration requests out on our first interface rather than an unbound interface, to avoid the problem of WACK replies being sent to the wrong port (w2k3 WINS server does this) (This used to be commit f7712ac7468184c07b3e3c10cb7b847ad1791dd2)
2007-10-10r5275: - added support for NBT_OPCODE_MULTI_HOME_REG (opcode 0xf) for WINS ↵Andrew Tridgell6-21/+228
name registrations - fixed a bug in the send queue handling on timeouts - added support for handling unexpected replies (replies to the wrong port) at the nbtsocket layer - added separate layer 2 code for wins refresh and wins registration (This used to be commit 2502b02898407e3262c09a5a4aa573c5f87b8f5f)
2007-10-10r5274: fixed some const warnings by making the str_list_ functions return ↵Andrew Tridgell1-7/+7
"const char **" (This used to be commit 4165f2163530c7d2ffbea2922fbfe942eee85e7a)
2007-10-10r5273: fixed another bug in the code that keeps timed eventsAndrew Tridgell1-2/+2
sorted. Hopefully it really works now :-) (This used to be commit 34e1a3ad7849df913584074ec406c2fcc40212fc)
2007-10-10r5266: This is a nice typo ;-)Günther Deschner1-1/+1
Guenther (This used to be commit 77b99c03b85827a1e70b0de41c8070e38af27374)
2007-10-10r5261: translate nbt rcode errors to NTSTATUS codesAndrew Tridgell4-0/+34
(This used to be commit 554d1b70e73faeb1f5ecf88f31c5810d86d76200)
2007-10-10r5260: - show an error message on nmblookup failureAndrew Tridgell3-11/+10
- always try to enable broadcast on nbt name sockets (this matches samba3 behaviour better) (This used to be commit 919bc14e7bbc04479cf11f7a7fd4c5e46616ef46)
2007-10-10r5259: make sure we give the ip of the interface that a name query comes inAndrew Tridgell3-8/+25
on as the first IP in a multi-homed reply (This used to be commit a9128f6544d56a637e28430cbd2907acbb978281)
2007-10-10r5256: More verbose description of functions (as I learn the code).Rafal Szczesniak1-5/+49
rafal (This used to be commit 38ac6b8d570fa02bc32899521d42ba7237f77024)
2007-10-10r5253: need to pre-declare some structuresAndrew Tridgell1-2/+3
(This used to be commit a4d44d4a9d7cf51a9a4e47dce8e603ba22324d99)
2007-10-10r5252: - fixed nmblookup for the nbt api changesAndrew Tridgell2-2/+72
- added a simple WINS server name registration and query test (This used to be commit d56e68ebf584e50aa409d22b1ca7c9276abea962)
2007-10-10r5251: - renamed the nbtd server side structures to have a nbtd_ prefix, toAndrew Tridgell11-64/+287
be consistent with the function names - added WINS client support to the NBT server. It will do initial WINS registration, and WINS refresh, automatically failing over to secondary WINS servers and handling multi-homed servers where we need to register multiple IPs. - added support for multi-homed name query replies, which are essential for multi-homed registration as the WINS server will query us to ensure we have the names when doing the secondary IPs in multi-homed registration (This used to be commit a1553fa8054dc7d33f5d77f8f95d3ffd90392b2a)
2007-10-10r5250: - added low level support for retrying nbt name queries, rather thanAndrew Tridgell5-63/+297
having the 2nd layer functions do retries themselves. This makes the code simpler, and allows the TRN_ID to be reused in the retry (which is how it is supposed to work). - added support for WACK replies to nbt name requests. A WACK reply specifies a timeout to wait for the real reply. - added WINS name refresh async calls, supporting multiple wins servers and multiple IPs to register (This used to be commit 76be35cb990de830c2451d9e48cb2c40a4befdb7)
2007-10-10r5249: don't include ';' in the default list separators for parsingAndrew Tridgell1-1/+1
smb.conf. It was rarely (if ever?) used, and poses problems as it is not multi-byte safe for character searches (This used to be commit 75a6cfb2f6b4ef85cf7e49958cccae9d35c4e5a2)
2007-10-10r5248: fixed a silly bug in DLIST_ADD_AFTER()Andrew Tridgell1-0/+1
(This used to be commit e36b6882fdd1bb331f297b3b82bc3621b8ba0586)
2007-10-10r5247: Fix rpcclient to work with new swig interface.Tim Potter1-18/+34
(This used to be commit ad12a90d0e0fe7106b1845097d8fc71be3b8e378)
2007-10-10r5245: Remove dead code.Tim Potter1-671/+1
(This used to be commit 949f31ad2865699da0f3a9e827606abd762b10b7)
2007-10-10r5244: Convert this module to use the new structure mapping SWIG stuff ↵Tim Potter1-72/+103
instead of the old dictionary based routines. The SAMR connect, close, enumdomains, opendomain, enum{users,aliases,groups} functions now work again! (This used to be commit 53b361657fe892ad7b9d781450ad841a6b1072e5)
2007-10-10r5243: Woohoo - memory leak city! Comment out talloc_free() call until ITim Potter1-1/+1
figure out a better way to pass tallocated memory around with SWIG. (This used to be commit c63668cbbe393a78ccd9a86a33fb4bd7bb9af16d)
2007-10-10r5242: Check that argument is an integer or a long for uint32_t inputTim Potter1-3/+16
typemap. The uint32_t output typemap must return a Python long as an unsigned uint32_t cannot be fully represented by a Python int. Likewise for the NTSTATUS typemap. (This used to be commit 6dba5d6fb43f87bedc380c4f82b1d5f90b70325a)
2007-10-10r5241: Generate swig wrappers for unions as well as structures.Tim Potter1-2/+2
(This used to be commit e65741159d8578d629ccfcd863f034fda895f9fe)
2007-10-10r5240: Don't return WERR_OK when no values were found (reported by Matt Cobb)Jelmer Vernooij1-4/+4
(This used to be commit f2e6d71584672e9bf0eaa7176d76781a30d317d1)
2007-10-10r5237: Add error code for "class not registered"Jelmer Vernooij2-0/+3
(This used to be commit b72a0ac654857273eaaf3c5e32d86abed0af3ceb)
2007-10-10r5224: Add in/out typemaps for resume handles. This saves us having to muchTim Potter1-0/+39
around with pointers to just one uint32. Add an output typemap to copy a policy handle as the talloc context is destroyed before the wrapper function returns. More work here needed to avoid memory leaks. Use the swig carrays.i file to create accessor and setter functions for fixed width integer types. Also add functions for struct samr_SamEntry as it's returned by the LookupDomain RPC. This really needs to be done by pidl so I don't have to go through and find all the structures that are returned in arrays. Include security.i to give us SIDs and security descriptors. (This used to be commit 5a1f6c999ef5e84e93ec8f07b9751d795c4566a5)
2007-10-10r5223: Rename dom_sid2 to dom_sid as we don't care about the differenceTim Potter1-1/+5
for the swig wrappers. (This used to be commit edb32b9b5175f720bce110bf2bb4cf136ed47142)
2007-10-10r5222: made the nbtd_self_packet() code more efficientAndrew Tridgell1-6/+16
(This used to be commit 8fe477955d0e28c891281803d625d80cb78a51b0)
2007-10-10r5221: replace the str_list_*() code with new code based on talloc(). This isAndrew Tridgell4-132/+104
a precursor to adding the wins client code in the nbt server. (This used to be commit e8e499755ab667015740b35a7787134ebe852954)
2007-10-10r5217: avoid epoll_ctl() if the event flags are already set correctlyAndrew Tridgell1-1/+4
(This used to be commit d89b493aaf690ceb107bd62df75f1bba82284f90)
2007-10-10r5216: don't defend group names against incoming name registration requestsAndrew Tridgell1-1/+1
(This used to be commit c5f5e1d4012c9f4fd4ea78a4276e6766727a5494)
2007-10-10r5215: register aliases as both client and server node types, so nmblookup ↵Andrew Tridgell1-0/+1
can see them (This used to be commit 3e9788691fe6d618b1e170494e818d0cbdc64eae)
2007-10-10r5214: added support for "netbios aliases" in smb.confAndrew Tridgell1-1/+8
(This used to be commit 3270b07539d9a50763f4f15c7aa8fee028928b29)
2007-10-10r5213: do our name broadcast refresh requests as register packets not refreshAndrew Tridgell1-4/+9
packets, as w2k3 and Samba3 do not defend against broadcast name refresh packets (This used to be commit 3935b5f7c57be46042110911ba5b00ac39b0f41a)
2007-10-10r5212: added checking for receiving our own packets as broadcastsAndrew Tridgell2-0/+35
(This used to be commit 290dbd8cdcfa6a897647768dcbbd03f22fc7cf36)
2007-10-10r5211: added broadcast name defense against both registration and refreshAndrew Tridgell4-7/+124
requests (This used to be commit 9eafe2cacaef64384febe6bb0938294f7c1ee6bf)
2007-10-10r5210: changed server side nbt functions to be prefixed with nbtd_ instead ofAndrew Tridgell8-76/+79
nbt_, so as to more clearly separate them from the client code in libcli/nbt/ (This used to be commit b07a7e35f26204055a99abf72438b5cd7ec35d3b)
2007-10-10r5209: Fix the endpoint mapper to work with IPX endpoints (whichJelmer Vernooij7-123/+160
accidently have the same protocol id as UUID's) Before this, Samba would give NDR errors when contacting a remote server that has IPX support enabled. This one was on my long due bugs list. (This used to be commit 7b847de64f35b8e897b64ad047d8aea3813214f8)
2007-10-10r5199: fix some minor configure bugsStefan Metzmacher3-3/+3
metze (This used to be commit 274ef2a206aa00b3155adc27f5b7e35d3fa52bf6)
2007-10-10r5198: don't consider failure to remove an epoll event as enough reason toAndrew Tridgell1-4/+2
fallback to select(). This can happen in too many situations. (This used to be commit 2d7242f2ec945c1e5fbc6f256265471f60ac8cd9)
2007-10-10r5197: moved events code to lib/events/ (suggestion from metze)Andrew Tridgell41-46/+46
(This used to be commit 7f54c8a339f36aa43c9340be70ab7f0067593ef2)
2007-10-10r5196: fixed sily bug (that metze found)Andrew Tridgell1-1/+1
(This used to be commit 180f29f9e05e75e8ed32d010bc0c57a55ea9d843)
2007-10-10r5195: most events don't need the time of the event, so save a ↵Andrew Tridgell16-39/+32
gettimeofday() call and just use timeval_current() when its actually needed (This used to be commit 236403cc4dc2924ed6a898acae0bb44cc1688dcc)
2007-10-10r5194: added support for using epoll instead of select() on systems that haveAndrew Tridgell2-48/+223
it. epoll is much more scalable than select(), but only exists on some systems (such as Linux with the 2.6.x kernel). The code detects any epoll system call failures at runtime and falls back to select() if there is a problem, so it should be safe to compile this on a 2.6 kernel and run it on a 2.4.x kernel. The speedup is quite large. It gains 20% in packet rate in the BENCH-NBT test, on top of another 20% gain from the better timer handling I added earlier. The really big gain will be when we are dealing with large numbers of file descriptors. With epoll we can handle hundreds of file descriptors all O(1), whereas with select it is O(n). (This used to be commit 26aa1aa69377e43da2942a42d137b95801e9fa1a)
2007-10-10r5193: make sure we mark the event dead when we free it on a dead connectionAndrew Tridgell2-1/+2
(This used to be commit 90535bab957ddaa7bfcdf43e199581f3352bdc8a)
2007-10-10r5190: Grr - typo.Tim Potter1-1/+1
(This used to be commit 5f40ff0f554323ed0a806be7a4d0f1fac97e3ffb)
2007-10-10r5189: fixed a double free bug in the ltdb indexing codeAndrew Tridgell1-7/+0
(This used to be commit 7be0bc93bd6757e52fd10bd3d3b3d1a8f5221452)
2007-10-10r5188: Add config.mk file for swig.Tim Potter1-0/+1
(This used to be commit c40fb6625d5cc6b3ddadfdc3c63d7856a45ec134)
2007-10-10r5187: ordered the timed events in the events code, which makes processingAndrew Tridgell2-95/+75
events much more efficient (no linked lists need to be traversed, so large numbers of timers are no problem) (This used to be commit b45b9436d78b0ad288b27a1023579bb566ca6202)